Who Belongs Together?
Group similar items together without labels.
What you'll learn in this level
- How an AI groups things with no labels at all
- That similarity is measured as a distance between numbers
- Why a person has to interpret the groups it finds
Groups With No Names
Sometimes a computer should put similar things into groups. And nobody tells it up front which groups there are.
It's like sorting a pile of mixed fruits into groups based on how they look, without someone telling you 'these are apples' and 'these are oranges.'
How grouping works
- 1
Finding Similarities
The computer looks for things with similar features. It puts those together.
- 2
No Labels Needed
When you sorted, the names were given. Here the computer needs none.
- 3
How People Use It
Streaming services group similar films and songs this way. Then they suggest new ones to you.
- 4
Discovering Patterns
This finds hidden patterns in data. People often miss them.
Here you group animals by their features.
Guess first, then experiment
The AI groups animals with no labels at all. Does it then know what its groups are called?